Critical illness poses an enormous burden on the health system in terms of morbidity, mortality, and costs. In order to reduce this burden many experts have proposed limiting the supply of intensive care unit (ICU) beds, forcing clinicians to limit ICU utilization through implicit rationing of services. The potential value of ICU bed limitations hinges on the econometric notion of demand elasticity, also known as supply induced demand, which holds that the availability of ICU beds in part influences their utilization There is substantial indirect evidence in support of demand elasticity in the ICU: in comparison to other similarly developed nations, the United States has more ICU beds per capita and admits more patients to the ICU without achieving better outcomes; and hospitals vary widely in the proportion of patients admitted to the ICU without differences and outcome. However, at present direct evidence is limited to small, single-center studies examining how ICU admission patterns change with changing ICU bed numbers. Although policy levers to control the ICU bed supply exist, such as certificate of need laws, more robust evidence is needed before we can consider these options. To fill this critical knowledge gap, we will perform the first multicenter, nationally representative study of the effect of changes in the ICU bed supply on intensive care utilization. Our central hypothesis is that as the ICU bed supply increases ICU case mix also changes in specific ways-low severity patients that are too well to benefit and high severity patients that are too sick to benefit will be more likely to be admitted to the ICU. To test this hypothesis we will capitalize on a natural experiment by which some US hospitals add ICU beds over time, whereas others do not. Using nationally representative data from Healthcare Cost and Utilization Project's (HCUP) State Inpatient Database (SID), we will apply a difference-in-differences approach that compares hospitals with and without an increasing ICU bed supply, controlling for other factors. First, we will characterize the effect of a change in the ICU bed supply on the proportion of ICU patients with relatively low illness severity, who are likely to survive without ICU admission. Second, we will determine the effect of a change in the ICU bed supply on the utilization of intensive care services among patients admitted to the hospital with terminal illness, for whom admission to the ICU unlikely to alter their survival. Together, these aims will provide the robust evidence necessary to support clinicians, health administrators, and policy makers in their efforts to develop local, regional, and national strategies to optimize resource utilization and improve patient outcomes. In addition, through a multifaceted research training plan, this project will provide the applicant with essential skills in econometric techniques, statistical modeling, and health policy analysis. Ultimately, these skills will facilitte his transition to a mentored career development award and a career as an independent critical care health services investigator capable of studying how system-level interventions affect critical care utilization and outcomes.